KR20240027756A - Point cloud attribute coding method, point cloud attribute decoding method, and terminal - Google Patents

Abstract

This application discloses a point cloud attribute coding method, a point cloud attribute decoding method, and a terminal, and belongs to the field of point cloud processing technology. The point cloud attribute coding method includes the steps of obtaining a point cloud to be coded; Splitting the point cloud to be coded into at least one coding point cloud block based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point; For each coding point cloud block containing at least two coding points, dividing the coding point cloud block into N coding point cloud groups based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order. Bar, N is a constant number; Obtaining a quantified first transform coefficient by performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group; and generating a target code stream by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group.

Description

Point cloud attribute coding method, point cloud attribute decoding method, and terminal

[Reference to related applications]

This application claims priority to Patent Application No. 202110970524.3 filed in China on August 23, 2021, the entire content of which is hereby incorporated by reference.

This disclosure relates to the field of point cloud processing technology, and particularly to a point cloud attribute coding method, a point cloud attribute decoding method, and a terminal.

A point cloud is a set of discrete points, irregularly distributed in space, representing the spatial structure and surface properties of a three-dimensional object or scenario.

Attribute coding of point clouds is divided into attribute prediction coding and attribute transformation coding. In the attribute prediction coding process, residual information corresponding to each coding point is obtained, and then quantification and entropy coding are performed on the corresponding true information to create a binary code. Acquire a stream; In the attribute transformation coding process, residual information corresponding to each coding point is obtained, and then transformation is performed on the residual information using a fixed-order matrix to obtain coefficients.

In the attribute prediction coding and attribute transformation coding processes, the correlation between residual information corresponding to each coding point is not considered, and redundant information exists in the residual information and transformation coefficients, thereby reducing the attribute coding efficiency of the point cloud. Additionally, since the point cloud attribute decoding process is consistent with the point cloud attribute coding process, it also reduces the point cloud attribute decoding efficiency.

An embodiment of the present application provides a point cloud attribute coding method, a point cloud attribute decoding method, and a terminal. In the attribute coding and attribute decoding process, the correlation between residual information corresponding to each coding point is not considered, and the attribute of the point cloud is not considered. Deterioration of coding/decoding efficiency can be solved.

In a first aspect, a point cloud attribute coding method is provided, the method comprising:

Obtaining a point cloud to be coded, ranking coding points in the point cloud to be coded according to a preset coding order;

Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. step;

For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. dividing, where N is an integer;

Transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a first quantified transform coefficient, and the first target coding point cloud group is the N coding point clouds. at least part of a group, wherein the first target coding point cloud group is a coding point cloud group that performs transform coding;

A target code stream is generated by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group, wherein the second target coding point cloud group is the N coding It includes the step of being a coding point cloud group excluding the first target coding point cloud group among the point cloud groups.

In a second aspect, a point cloud attribute decoding method is provided, the method comprising:

Performing decoding on a code stream to be decoded to obtain a point cloud to be decoded, wherein decoding points in the point cloud to be decoded are ranked according to a preset coding order;

Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. step;

For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the decoding point cloud block and the preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. dividing, where M is an integer;

Inverse quantification processing is performed on the second transform coefficient corresponding to the third target decoding point cloud group to obtain target high-frequency coefficients and target low-frequency coefficients, wherein the third target decoding point cloud group is the M decoding point cloud group. wherein the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing;

and performing inverse transformation on the target high-frequency coefficient and the target low-frequency coefficient to obtain third residual information corresponding to the third target decoding point cloud group.

In a third aspect, providing a coder,

A first acquisition module that acquires a point cloud to be coded, where coding points in the point cloud to be coded are ranked according to a preset coding order;

Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. first division module;

For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. A second division module where N is an integer for division;

Transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a first quantified transform coefficient, and the first target coding point cloud group is the N coding point clouds. a first coding module that is at least part of a group, and wherein the first target coding point cloud group is a coding point cloud group that performs transform coding;

A target code stream is generated by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group, wherein the second target coding point cloud group is the N coding It includes a second coding module that is a coding point cloud group excluding the first target coding point cloud group among the point cloud groups.

In a fourth aspect, providing a decoder,

a decoding module that performs decoding on a code stream to be decoded to obtain a point cloud to be decoded, wherein decoding points in the point cloud to be decoded are ranked according to a preset coding order;

Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. third split module;

For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the decoding point cloud block and the preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. A fourth division module where M is a constant number for division;

Inverse quantification processing is performed on the second transform coefficient corresponding to the third target decoding point cloud group to obtain a target high-frequency coefficient and a target low-frequency coefficient, wherein the target decoding point cloud group is at least one of the M decoding point cloud groups. an inverse quantification module, wherein the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing;

and an inverse transformation module that performs inverse transformation on the target high-frequency coefficient and the target low-frequency coefficient to obtain third residual information corresponding to the third target decoding point cloud group.

In a fifth aspect, a terminal is provided, wherein the communication device includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, and when the program or instruction is executed by the processor, Implement the steps of the point cloud attribute coding method of the first aspect, or implement the steps of the point cloud attribute decoding method of the second aspect.

In a sixth aspect, a readable storage medium is provided, wherein a program or instruction is stored in the readable storage medium, and when the program or instruction is executed by a processor, the steps of the point cloud attribute coding method of the first aspect are: Implements, or implements the steps of the point cloud attribute decoding method of the second aspect.

In a seventh aspect, a chip is provided, wherein the chip includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor executes a program or instruction, and the point cloud of the first aspect is provided. Implement steps of the attribute coding method, or implement steps of the point cloud attribute decoding method of the second aspect.

In an eighth aspect, there is provided a computer program/program product, wherein the computer program/program product is stored in a non-volatile storage medium, and the computer program/program product is executed by at least one processor. Implement steps of the point cloud attribute coding method, or implement steps of the point cloud attribute decoding method of the second aspect.

In a ninth aspect, there is provided a communication device, configured to implement the steps of the point cloud attribute coding method of the first aspect, or to implement the steps of the point cloud attribute decoding method of the second aspect.

In an embodiment of the present application, after obtaining a point cloud to be coded, the point cloud to be coded is divided into at least one coding point cloud block, point cloud grouping is performed on the coding points in each coding point cloud block, and coding is performed. Split the point cloud block into N coded point cloud groups. In the transform coding process, the correlation between residual information corresponding to each coding point is sufficiently considered, and transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient. is obtained, where there is a correlation between the residual information of each coding point in the first target coding point cloud group. Furthermore, entropy coding is performed on the quantified first transform coefficient and the second residual information corresponding to the second target coding point cloud group to generate a target code stream. The coding process sufficiently considers the correlation between residual information corresponding to each coding point, performs transformation coding on the first residual information corresponding to the first target coding point cloud group, and reduces redundant information in the transformation coefficients to form a point cloud. The efficiency of attribute coding has been improved.

1 is a diagram of the point cloud AVS encoder framework.
Figure 2 is a diagram of the point cloud AVS decoder framework.
Figure 3 is a flow chart of conventional point cloud attribute coding.
Figure 4 is a flowchart of a point cloud attribute coding method provided by an embodiment of the present application.
Figure 5 is a flowchart of a point cloud attribute decoding method provided by an embodiment of the present application.
Figure 6 is a structural diagram of an encode provided by an embodiment of the present application.
Figure 7 is a structural diagram of a decoder provided by an embodiment of the present application.
Figure 8 is a structural diagram of a communication device provided by an embodiment of the present application.
Figure 9 is a hardware structure diagram of a terminal provided by an embodiment of the present application.

A clear explanation of the technical solutions in the embodiments of this application will be provided with reference to the drawings in the embodiments of this application below. It is obvious that the described embodiments are only some embodiments of the present application and are not all embodiments. will be. All other embodiments acquired by those skilled in the art based on the embodiments of the present application should be considered to fall within the scope of the present application.

The terms "first", "second", etc. in the specification and claims of this application are only used to distinguish similar objects and are not intended to describe a specific order or sequential order. The terms so used are interchangeable under appropriate circumstances, so that embodiments of the present application may be practiced in a different order than shown or described herein, and the objects distinguished by "first", "second", etc. It will be understood that they are generally of the same type, and the quantity of objects is not limited; for example, the first object may be one or plural. In addition, "and/or" used in the specification and claims indicates at least one of the connected objects, and the symbol "/" generally indicates that the preceding and related objects are in the relationship of "or".

The encoder corresponding to the point cloud attribution method and the decoder corresponding to the cloud attribution method of the embodiments of the present application may both be terminals, and the terminal may also be referred to as a terminal device or user equipment (User Equipment, UE), and the terminal may be Cell phone, Tablet Personal Computer, Laptop Computer or Notebook Computer, Personal Digital Assistant (PDA), Palmtop, Netbook, Ultra-mobile Personal Computer (UMPC), Personal Portable Personal Digital Assistant (PDA), Mobile Internet Device (MID), augmented reality (AR)/virtual reality (VR) devices, robots, wearable devices, vehicles It may be a terminal-side device such as Vehicle User Equipment (VUE) or Pedestrian User Equipment (PUE), and wearable devices include smart watches, bands, earphones, glasses, etc. It should be noted that the embodiments of the present application do not limit the specific type of terminal.

To facilitate understanding, some details mentioned in the examples of this application will be explained below.

Referring to Figure 1, as shown in Figure 1, in the current digital audio and video coding and decoding technology standards, the point cloud source coding standard (Audio Video coding Standard, AVS) coder is used to obtain the geometric information and properties of the point cloud. Information is classified and coded. First, the geometric information is converted to coordinates so that the point cloud is all contained within the bounding box, and then coordinate quantification is performed. Quantification mainly plays the role of shrinkage. Because quantification rounds the geometric coordinates, the geometric information of some points is the same, so they are called duplicate points, and determines whether to remove duplicate points according to the parameters. Quantification and duplicate point removal are two of them. The step is also called the voxelization process. Next, split the bounding box into multitrees such as octrees, quadtrees, or binary trees. In the multitree-based geometric information coding framework, the bounding box is divided into 8 subcubes, and the non-empty subcubes are continuously divided, stopping the division when the lift node is a unit cube of 1x1x1, and the point of the leaf node is Code the numbers to create a binary code stream.

When performing multi-tree based geometric coding on a point cloud, the point to be coded must store the occupancy information of adjacent nodes and perform predictive coding on the occupancy information of the point to be coded, so the point to be coded close to the leaf node In other words, a large amount of occupancy information must be stored, which occupies a large amount of memory space.

Once geometric coding is completed, the geometric information is reconstructed for later re-coloring. The main goals of attribute coding are color and reflectance information. First, it is determined whether to perform color space conversion according to the parameters, and if color space conversion is performed, color information is converted from RGB (Red Green Blue) color space to luminance color (YUV) space. Then, the geometrically reconstructed point cloud is re-colored using the raw point cloud, so that the uncoded attribute information corresponds to the reconstructed geometric information. In color information coding, the point cloud is ranked by Morton code or Hilbert code, then the closest reconstructed attribute value of the point to be predicted is found using geometric spatial relationships, and the predicted point is predicted using the found adjacent reconstructed attribute value to obtain the predicted attribute value. After obtaining, the prediction residual is obtained by differentiating the actual attribute value from the predicted attribute value, and finally, the prediction residual is quantified and coded to generate a binary code stream.

The decoding process in the digital audio and video coding and decoding technology standards corresponds to the coding process described above, and specifically, the AVS decoder framework is as shown in FIG. 2.

Referring to FIG. 3, FIG. 3 is a flowchart of conventional point cloud attribute coding. The existing way to perform attribute coding on point clouds is as follows.

First, the prediction residual is obtained by subtracting the prediction signal from the original signal, and the prediction residual is also called residual information. Discrete Cosine Transform (DCT) is performed on the residual information to obtain a first-order transform coefficient, and the first-order transform coefficient can be displayed as a first-order transform coefficient block; A secondary transformation coefficient with more concentrated energy is obtained by performing secondary transformation on the low-frequency component of the preset area of the corresponding primary transformation coefficient block, where, as shown in FIG. 3, the corresponding preset area is 1 It is located in the upper left corner of the difference conversion coefficient block. The corresponding quadratic transformation coefficient can be expressed as a quadratic transformation coefficient block; A binary code stream is obtained by performing quantification and entropy coding on the low-frequency components of the preset area of the corresponding secondary transform coefficient block.

In addition, a reverse secondary transformation is performed on the quantified secondary transformation coefficient to obtain a primary transformation coefficient, and a reverse primary transformation is performed on the primary transformation coefficient to obtain residual information. Furthermore, reconstruction information is obtained by performing an addition operation on the residual information and predicted attribute values, and loop filtering is performed on the reconstruction information to obtain attribute reconstruction values.

Here, in some specific coding scenarios, the secondary transform and reverse secondary transform are not performed.

Current general-purpose point cloud technology standards have the following technical problems:

In the point cloud attribute coding process, the correlation between residual information corresponding to each coding point is not sufficiently considered, so there is still some redundant information in the transformation coefficient, and the attribute coding efficiency of the point cloud is relatively low.

Based on the above situation, how can in the process of point cloud attribute coding and attribute decoding, fully utilize the correlation between the residual information corresponding to each coding point to perform attribute coding and attribute decoding to improve point cloud attribute coding and point cloud attribute decoding efficiency? Whether to improve it is a technical challenge that must be solved.

Referring to the drawings below, a detailed description will be given of the point cloud attribute coding method provided by the embodiments of the present application through some embodiments and application scenarios.

Referring to FIG. 4, FIG. 4 is a flowchart of the point cloud attribute coding method provided by the present application. The point cloud attribute coding method provided by this embodiment includes the following steps.

S101: Obtain a point cloud to code.

The point cloud to be coded during this step is a point cloud that has undergone recoloring and color space conversion, and the recoloring is a geometric reconstruction using the original point cloud. Recoloring is performed on the point cloud, and the uncoded attribute information is reconstructed into a geometric point cloud. Obtain a recolored point cloud by matching it with information; The color space conversion refers to converting the color information of the point cloud from RGB space to YUV space.

Here, the point cloud to be coded can be understood as a point cloud sequence, and each coding point in the point cloud to be coded can be understood as each element in the point cloud sequence. Points to be coded in the point cloud sequence are ranked according to a preset coding order. An optional implementation method is to perform Hilbert ranking on the points to be coded in the point cloud sequence, and each point to be coded is small according to the Hilbert code. They are ranked from small to large, or Morton ranking is performed on the points to be coded in the point cloud sequence, and they are ranked from small to large according to the Morton code.

S102: Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, the point cloud to be coded is divided into at least one coding point cloud block. Divide.

In this step, segmentation is performed on the point cloud to be coded to obtain a coded point cloud block. Here, the division method for the point cloud to be coded is determined based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each coding point. Specific technology The scheme may refer to subsequent embodiments.

In one selectable implementation method, the number of coding point cloud blocks is 1, and in this case, it can be understood that the entire point cloud to be coded is regarded as one coding point cloud block.

Since the attribute information corresponding to the coding points belonging to the same coding point cloud block has spatial correlation, the corresponding attribute information also has a stronger correlation, that is, the geometric positions of these coding points are closer and the color attributes are more similar. You must understand.

If the coding points in the point cloud sequence are ranked according to Hilbert code from smallest to largest, then after performing segmentation on the point cloud to be coded, it should be understood that the Hilbert codes corresponding to the coding points belonging to the same coding point cloud block are the same. do.

If the coding points in the point cloud to be coded are ranked according to the Morton code from smallest to largest, after performing segmentation on the point cloud to be coded, determine that the Morton codes corresponding to the coding points belonging to the same coding point cloud block are the same. You must understand.

S103: For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point clouds. Divide into groups.

In this step, the point cloud to be coded is divided into at least one coding point cloud block, and then, for the coding point cloud block containing at least two coding points, grouping is performed on the corresponding coding point cloud block to create a coding point cloud block. The block is divided into N coding point cloud groups, where N is an integer. It should be understood that point cloud grouping can be performed on the coding point cloud block using the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, and specific technical solutions can be referred to subsequent embodiments. there is.

In a possible case, the number of coding point cloud groups is one, and in this case, it can be understood that the entire coding point cloud block is regarded as one coding point cloud group.

After dividing the coding point cloud block into N coding point cloud groups, for any coding point in the coding point cloud group, search for the nearest neighbor of the coding point using geometric spatial relationships, and reconstruct the found neighbor. After performing prediction on the corresponding coding point using the attribute value to obtain the predicted attribute value, the difference between the actual attribute value and the predicted attribute value of the coding point is performed to obtain the prediction residual. The prediction residual is It is also called residual information. In this way, residual information corresponding to each coding point among each coding point cloud group is obtained.

S104: Perform transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient.

In this step, all coding point cloud groups in the coding point cloud blocks may be determined as the first target coding point cloud group, or some coding point cloud groups in the coding point cloud blocks may be determined as the first target coding point cloud group, For technical solutions on how to specifically determine the target coding point cloud group, refer to subsequent embodiments.

As described above, after point cloud grouping is performed on the coding point cloud block, residual information corresponding to each coding point among each coding point cloud group is determined. In this step, a set of residual information corresponding to each coding point among the first target point cloud group may be determined as residual information corresponding to the first target coding point cloud group.

For specific technical solutions on how to obtain a quantified first transform coefficient by performing transform coding and quantification processing on the residual information corresponding to the first target coding point cloud group, refer to subsequent embodiments.

S105: Entropy coding is performed on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group to generate a target code stream.

In this step, after obtaining residual information corresponding to the first transform coefficient and the second target coding point cloud group, entropy coding is performed on the first transform coefficient and the first residual information to generate a target code stream.

It should be understood that the second coding target point cloud group is a coding point cloud group excluding the first target coding point cloud group among N coding point cloud groups.

Here, entropy coding is a coding method in which no information is lost during the coding process according to the principle of entropy, and the entropy coding may be Shannon coding, Huffman coding, or other types of coding, and the present embodiment is specifically limited to this. I never do that. Here, the target code stream is a binary code stream.

It should be understood that in some embodiments, inverse quantification and inverse transformation operations may be performed on the first transformation coefficients to obtain attribute reconstruction values.

In an embodiment of the present application, after obtaining a point cloud to be coded, the point cloud to be coded is divided into at least one coding point cloud block, point cloud grouping is performed on the coding points in each coding point cloud block, and coding is performed. Split the point cloud block into N coded point cloud groups. In the transform coding process, the correlation between residual information corresponding to each coding point is sufficiently considered, and transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient. Obtaining: wherein there is a correlation between the residual information of each coding point in the first target coding point cloud group; Furthermore, entropy coding is performed on the quantified first transform coefficient and the second residual information corresponding to the second target coding point cloud group to generate a target code stream. The coding process sufficiently considers the correlation between residual information corresponding to each coding point, performs transform coding on the first residual information corresponding to the first target coding point cloud group, and reduces transform coefficient redundancy information to reduce the properties of the point cloud. Coding efficiency has been improved.

In order to help understand the technical effects of the present application, the point cloud attribute coding method provided by the embodiment of the present application is applied to the PCEM platform to examine the attribute coding efficiency of the point cloud. Please refer to Table 1 for specific test results.

sequence BD-AttrRate
(Luma) BD-GeomRate 1
(Chroma Cb) BD-GeomRate 2
(Chroma Cr) sequence 1 -4.4% -6.2% -8.0% sequence 2 -4.8% -2.8% -3.2% sequence 3 -3.3% -9.9% -9.3% sequence 4 -2.6% -13.3% -14.7% Sequence 5 -2.3% -10.3% -9.7% Sequence 6 -3.4% -25.2% -26.6% Sequence 7 -1.1% -5.8% 0.9% sequence 8 -21.3% -54.0% -62.1% sequence 9 -6.3% -21.3% -30.9% Sequence 10 -6.1% -23.9% -29.2% Sequence 11 -6.2% -18.8% -23.1% Sequence 12 -4.9% -20.8% -23.9%

Here, BD-AttrRate in Table 1 represents the compression performance of the Y component of the transform coefficient, BD-GeomRate 1 represents the compression performance of the U component of the transform coefficient, and BD-GeomRate 2 represents the compression performance of the V component of the transform coefficient. indicates. The sequence in Table 1 is a code stream sequence after comparing the point cloud attribute coding method provided by the embodiment of the present invention with the existing point cloud attribute coding method, and the BD-AttrRate, BD-GeomRate 1, and BD-GeomRate 2 are negative numbers. This indicates that performance is improving, and based on this, it should be understood that the larger the absolute value of the BD-rate, the greater the performance gain.

As can be seen in Table 1, the coding performance of point cloud attribute coding can be improved by using the point cloud attribute coding method provided by the embodiment of the present application.

Optionally, the step of dividing the point cloud to be coded into at least one coding point cloud block based on the quantity of coding points included in the point cloud to be coded and the order of each coding point,

Using the first target shift value, the ranking code corresponding to each coding point is shifted to obtain the first target ranking code corresponding to each coding point, and the target shift value is stored in the point cloud to be coded. Determining based on the included coding point quantity and volume information of the bounding box;

and dividing coding points with the same first target ranking code into the same coding point cloud block.

In this embodiment, coordinate conversion is performed on the geometric information of the point cloud to be coded so that the point cloud is all contained within the bounding box. The bounding box can be understood as a cube covering the point cloud to be coded, and each corner of the cube is located on the X-axis, Y-axis, or Z-axis of the coordinate system. A third shift parameter corresponding to the point cloud to be coded is determined based on the size of the corresponding bounding box, that is, the side length of the corresponding bounding box in the X-axis, Y-axis, and Z-axis of the coordinate system.

The quantity of coding points included in the point cloud to be coded is obtained, and a fourth shift parameter corresponding to the point cloud to be coded is determined based on the quantity of coding points included in the coding point cloud.

In this embodiment, a shift calculation formula is also set in advance, and the third shift parameter and the fourth shift parameter are input into the shift calculation formula to obtain a first target shift value corresponding to the point cloud to be coded. Here, the third shift parameter can be obtained by inputting the side length corresponding to the bounding box into the third shift parameter calculation formula; The fourth shift parameter can be obtained by inputting the quantity of coding points included in the point cloud to be coded into the fourth shift parameter calculation formula.

Specifically, the shift calculation formula is consistent with the shift calculation formula of the subsequent point cloud attribute decoding method, the third shift parameter calculation formula is consistent with the third shift parameter calculation formula of the subsequent point cloud attribute decoding method, and the fourth The shift parameter calculation formula is consistent with the fourth shift parameter calculation formula of the subsequent point cloud attribute decoding method.

Below, we will explain as an example that the coding points in the point cloud to be coded are ranked according to Hilbert code from smallest to largest.

As described above, coding points in the point cloud to be coded are ranked according to Hilbert code from smallest to largest. After obtaining the first target shift value, the Hilbert code corresponding to each coding point is shifted to the right using the first target shift value to obtain a new Hilbert code corresponding to each coding point, and the new Hilbert code is It is decided by the target Hilbert code, and the target Hilbert code can be understood as a target ranking code. The first target shift value represents the number of bits shifted during the shifting process of the coding point, and the first target shift value may also represent the density of the point cloud to be coded. The smaller the target shift value, the more the number of bits in the point cloud to be coded. It is important to understand that the density of each coding point is high.

Furthermore, the cloud of coding points to be coded is divided according to the target Hilbert code corresponding to each coding point, and coding points with the same target Hilbert code are divided into the same coding point cloud block. In other words, the cloud of coding points to be coded is divided into the same coding point cloud block. The target Hilbert code corresponding to each coding point is the same.

In another embodiment, the first target shift value may be set to user-defined, or the first target shift value may be determined based on a geometric quantification step of the coding point cloud.

In this embodiment, a shift is performed on the Hilbert code of each coding point in the point cloud to be coded, and the point cloud to be coded is divided into coding point cloud blocks according to the target Hilbert code corresponding to each shifted coding point. Here, since the target Hilbert code of the coding points belonging to the same coding point cloud block is the same, it is considered to have a stronger correlation between the attribute information of these points, and the coding points whose attribute information has spatial correlation are included in the same coding point cloud block. Divide.

Below, we will describe in detail a technical method for performing point cloud grouping for the coding point cloud block using the quantity of coding points included in the coding point cloud block and the preset maximum transformation order.

In this embodiment, the maximum transform order is preset, and the preset maximum transform order is intended to indicate the maximum order corresponding to the transform matrix used in the transform coding process. Obtain the quantity of coding points included in the coding point cloud block, and if the quantity of coding points is less than or equal to the preset maximum transformation order, the coding point cloud block is determined as one coding point cloud group, that is, the corresponding coding point cloud block is obtained. Point cloud grouping is not performed on coding point cloud blocks. Here, a preset maximum conversion order can be recorded in the code stream.

If the corresponding coding point quantity is greater than the preset maximum transformation order, based on the corresponding coding point quantity and the first preset value, point cloud grouping is performed on the coding point cloud block, and the coding point cloud block is divided into at least two Divide the coding point cloud into groups. Here, based on the quantity of coding points included in the coding point cloud block and the first preset value, the specific method of performing point cloud grouping for the coding point cloud block may refer to subsequent embodiments.

Optionally, the step of dividing the coding point cloud block into at least two coding point cloud groups based on the coding point quantity included in the coding point cloud block and a first preset value,

If the third numerical value is greater than the preset maximum transformation order, performing a grouping calculation operation on the third numerical value using the first preset numerical value to obtain a fourth numerical value;

According to the preset coding order of coding points among the coding point cloud blocks, at least some coding points among the coding points for which point cloud grouping has not been performed until all coding points of all coding point cloud blocks complete point cloud grouping. It includes the step of dividing into one coding point cloud group.

A third numerical value is used to indicate the coding point quantity without performing point cloud grouping of the coding point cloud block, and if the third numerical value is greater than the preset maximum transformation order, the first preset numerical value is used for the third numerical value. A grouping calculation operation is performed to obtain a fourth value, where the fourth value is smaller than the preset maximum transformation order. Here, the specific method of how to use the first preset numerical value to perform a grouping calculation operation on the third numerical value to obtain the fourth numerical value may refer to subsequent embodiments.

After obtaining the fourth numerical value, at least some coding points among coding points for which point cloud grouping has not been performed are divided into one coding point cloud group according to a preset coding order of coding points among coding point cloud blocks. In other words, determine some coding points among coding point cloud blocks that have not performed point cloud grouping, and divide these some coding points into one coding point cloud group, or all coding points that have not performed point cloud grouping among coding point cloud blocks. Determine coding points and divide these coding points into one coding point cloud group. Here, the quantity of coding points divided into the at least one coding point cloud group is equal to the second value.

After performing point cloud grouping for some coding points, update the third value, and if the updated third value is still greater than the preset maximum transformation order, all coding points in the coding point cloud block complete the point cloud grouping. Until this happens, point cloud grouping is continuously performed on at least some coding points among the coding point cloud blocks.

Below, it will be described in detail how to obtain the fourth numerical value by performing a grouping calculation operation on the third numerical value using the first preset numerical value.

The process of the grouping calculation operation is as follows: performing a division operation on a third numerical value and a first preset numerical value to obtain a division result; Obtain the target number by rounding the division result. When the target value is greater than the preset maximum conversion order, update the third value to the target value, and continue for the third value and the first preset value until the target value is less than or equal to the preset maximum conversion order. Perform division association to determine the target number as the fourth number.

It should be understood that the technical method for performing point cloud grouping on coding point cloud blocks in this embodiment is consistent with the subsequent technical method for performing point cloud grouping on decoding point cloud blocks, and will not be described repeatedly here.

Optionally, before performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient, the method includes:

Obtaining a first transformation ID in the point cloud to be coded;

When the first transformation ID indicates that transformation coding is performed on all coding point cloud groups, determining all coding point cloud groups as the first target coding point cloud group;

When the first transformation ID indicates that self-adaptive transformation coding is performed on all coding point cloud groups, at least one coding point cloud group among the N coding point cloud groups is determined as the first target coding point cloud group. It includes steps to:

It should be understood that the first target coding point cloud grouping refers to a point cloud grouping that performs transform coding, that is, transform coding must be performed on each coding point in the first target coding point cloud group.

In this embodiment, the point cloud to be coded has a first transformation ID set in advance, and the first transformation ID is also called a self-adaptive transformation ID, and can be expressed as a self-adaptive transformation (AdpTransform), and the first transformation ID is can be set to self-definition.

If AdpTransform=0, it indicates that transformation coding is performed on all coding points, that is, transformation coding is performed on all coding point cloud groups. In this case, all coding point cloud groups are converted to the first target coding point cloud group. decide

If AdpTransform=1, it indicates that each coding point cloud group performs its own adaptive transformation coding, and if so, it is decided whether to perform transformation coding for that coding point cloud group according to the actual situation of each coding point cloud group. , In other words, at least one coding point cloud group among the N coding point cloud groups is determined as the first target coding point cloud group.

It should be understood that each coding point cloud group among the coding point cloud blocks is ranked according to the grouping time order of the point cloud group, and that the coding point cloud group ranked most forward among the coding point cloud blocks is determined as the third point cloud group.

In this embodiment, the fourth transformation ID is preset in the coding point cloud block, and the fourth transformation ID is used to convert the third point cloud group into the first target coding point when transformation coding is performed on the third point cloud group. This is to indicate that decisions are made by cloud group. In other words, the fourth transformation ID is preset in the coding point cloud block, and it is determined whether the third point cloud group is the first target coding point cloud group according to the fourth transformation ID.

Each coding point cloud group among the coding point cloud blocks is ranked according to the grouping time order of the point cloud group, and the coding point cloud group excluding the coding point cloud group ranked most forward among the coding point cloud blocks is determined as the fourth point cloud group. You must understand what you are doing. For any fourth point cloud group, the coding point cloud group adjacent to the fourth point cloud group among the coding point cloud blocks and located in front of the fourth point cloud group is determined as the adjacent point cloud group, and the adjacent point cloud group is determined. The maximum value among the reconstruction information corresponding to the cloud group is determined as the third reconstruction value, and the minimum value among the reconstruction information corresponding to the adjacent point cloud group is determined as the fourth reconstruction value.

Calculate the difference result between the third reconstruction value and the fourth reconstruction value, and obtain the absolute value of the difference result. Furthermore, in this embodiment, a second preset value is also set, and when the absolute value is smaller than the second preset value, the fourth point cloud group is determined as the first target decoding point cloud group.

Optionally, performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient, comprising:

calculating a multiplication result between a transformation matrix and the first residual information, and determining the multiplication result as a first transformation coefficient to be quantified;

and performing quantification on the first conversion coefficient to be quantified to obtain the quantified first conversion coefficient.

The first residual information is residual information corresponding to the first target coding point cloud group, and as described above, the residual information corresponding to the coding point cloud group is a set of residual information corresponding to each coding point of the coding group, , the first residual information can be understood as a one-dimensional residual information matrix.

The transformation matrix may be a Discrete Cosine Transform (DCT) matrix or a Discrete Sine Transform (DST) matrix or a Hadamard Transform matrix or another type of transformation matrix. The transformation order of the transformation matrix is the same as the number of code points included in the first target coding point cloud group. For example, if the number of code points included in the first target coding point cloud group is 3, the transformation order of the transformation matrix is The number is also 3, and the transformation matrix is a 3*3 matrix.

The quantified first conversion coefficient includes a low-frequency coefficient and a high-frequency coefficient, where the low-frequency coefficient is also called a DC coefficient, and the high-frequency coefficient is also called an AC coefficient.

A transformation coefficient to be quantified is obtained by multiplying the transformation matrix and the residual information corresponding to the first target coding point cloud group. For example, the following formula may be referred to.

Here, DC in the above formula represents the low-frequency coefficient, AC represents the high-frequency coefficient, AttrRes represents the first residual information, and T represents the element in the transformation matrix. From the above formula, if the first target coding point cloud group includes K coding points, the first residual information matrix includes K residual information, and the first transformation coefficient to be quantified is one low-frequency coefficient, K-1 It can be seen that it contains high-frequency coefficients, and the transformation order of the transformation matrix is K.

The color attribute of the first conversion coefficient to be quantified is the YUV color space, and the U and V components in the YUV color space can be determined as the first component, and the Y component can be determined as the second component.

Quantification is performed using a preset first quantification step for the high frequency coefficients in the U component and the V component to obtain a quantified third high frequency coefficient.

Quantification is performed using a preset second quantification step for the low-frequency coefficients in the U component and the V component to obtain a quantified third low-frequency coefficient.

Quantification is performed using distributed preset third quantification steps for the high-frequency coefficients and low-frequency coefficients in the Y component, to obtain quantified fourth high-frequency coefficients and fourth low-frequency coefficients.

In this embodiment, the quantified third high-frequency coefficient, the quantified third low-frequency coefficient, the quantified fourth high-frequency coefficient and the quantified fourth low-frequency coefficient are determined as the quantified first conversion coefficient.

Referring to the drawings below, a detailed description will be given of the point cloud attribute decoding method provided by the embodiments of the present application through some embodiments and application scenarios.

Referring to FIG. 5, FIG. 5 is a flowchart of the point cloud attribute decoding method provided by the present application. The point cloud attribute decoding method provided by this embodiment includes the following steps.

S201: Perform decoding on the code stream to be decoded to obtain a point cloud to be decoded.

In this step, after obtaining the code stream to be decoded, decoding is performed on the code stream to be decoded to obtain a point cloud to be decoded, the point cloud to be decoded includes a plurality of decoding points, and the point cloud to be decoded is obtained. The decoding points are ranked according to a preset coding order. Here, the decoding points of the point cloud to be decoded may be ranked according to the Hilbert code from small to large, or according to the Morton code from small to large.

S202: Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, the point cloud to be decoded is divided into at least one decoding point cloud block. Divide.

In this step, segmentation is performed on the point cloud to be decoded to obtain a decoding point cloud block. Here, the division method for the point cloud to be decoded is determined based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point. Specific technology The scheme may refer to subsequent embodiments.

It should be understood that attribute information corresponding to decoding points belonging to the same decoding point cloud block has spatial correlation, that is, the geometric positions of these decoding points are closer and the color attributes are more similar.

If the points to be decoded in the point cloud to be decoded are ranked according to Hilbert code from smallest to largest, then after performing segmentation on the point cloud to be decoded, the Hilbert codes corresponding to the decoding points belonging to the same decoding point cloud block are the same. You must understand that

If the decoding points in the point cloud to be decoded are ranked according to the Morton code from smallest to largest, after performing segmentation on the point cloud to be decoded, it is determined that the Morton codes corresponding to the decoding points belonging to the same decoding point cloud block are the same. You must understand.

S203: For each decoding point cloud block including at least two decoding points, based on the quantity of decoding points included in the point cloud block to be decoded and a preset maximum transformation order, the decoding point cloud block is divided into M decoding points. Divide into cloud groups.

In this step, the point cloud to be decoded is divided into at least one decoding point cloud block, and then, for the coding point cloud block containing at least two decoding points, grouping is performed on the corresponding decoding point cloud block to decode the point cloud. The block is divided into M decoding point cloud groups. Here, the implementation method of dividing the decoding point cloud block into M decoding point cloud groups may refer to the embodiment.

In one possible case, the decoding point cloud group quantity is 1, and in this case, it can be understood that the entire decoding point cloud block is regarded as one decoding point cloud group.

After dividing the decoding point cloud block into M decoding point cloud groups, for any one decoding point in the decoding point cloud group, search for the nearest neighbor of the decoding point using geometric spatial relationships, and reconstruct the found neighbor. By performing prediction on the corresponding decoding point using the attribute value to obtain the prediction attribute value, the prediction attribute value corresponding to each decoding point among each decoding point cloud group can be obtained, and the prediction attribute value can also be used as prediction information. You must understand that it is also called.

S204: Perform inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain the target high-frequency coefficient and the target low-frequency coefficient.

The third target decoding point cloud group is at least a part of the M decoding point cloud groups, and the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing.

In this step, it should be understood that at least some decoding point cloud groups among the M decoding point cloud groups are determined as the third target decoding point cloud group, and the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing. And, for technical solutions on how to specifically determine the target decoding point cloud group, refer to subsequent embodiments.

In this embodiment, the second transform coefficient corresponding to the third target coding point cloud group is read, and inverse quantification processing is performed on the second transform coefficient to obtain the target high-frequency coefficient and the target low-frequency coefficient. Here, the specific implementation method of the reverse quantification process may refer to the subsequent examples.

S205: Perform inverse transformation on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group.

In this embodiment, after obtaining the target high-frequency coefficient and the target low-frequency coefficient, inverse transformation processing is performed on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group.

In an optional implementation method, among the M decoding point cloud groups, a decoding point cloud group that does not perform inverse quantification processing is determined as the fourth target decoding point cloud group, and inverse quantification is performed on the fourth target decoding point cloud group. Thus, residual information corresponding to the fourth target decoding point cloud group can be obtained.

As described above, after dividing the decoding point cloud block into M decoding point cloud groups, for any one decoding point in the decoding point cloud group, search for the nearest neighbor of the decoding point using geometric spatial relationships, , prediction information corresponding to the coding point can be obtained by performing prediction on the corresponding decoding point using the reconstructed attribute value of the found neighbor. Furthermore, after determining the residual information corresponding to the coding point, the residual information corresponding to the coding point and the prediction information corresponding to the coding point are added to obtain reconstructed attribute information corresponding to the coding point.

The point cloud attribute decoding method provided by the embodiment of the present application sufficiently considers the correlation between residual information corresponding to each decoding point and performs inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group. Thus, the target high-frequency coefficient and the target low-frequency coefficient are obtained, where the attribute information of each decoding point in the third target decoding point cloud group has spatial correlation. Furthermore, inverse transformation processing is performed on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group, thereby reducing redundant information among the transform coefficient and residual information to decode the properties of the point cloud. improves efficiency.

Optionally, based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, the point cloud to be decoded is divided into at least one decoding point cloud block. The above step of dividing into,

Using the second target shift value, performing a shift on the ranking code corresponding to each decoding point to obtain a second target ranking code corresponding to each decoding point;

and partitioning decoding points with the same second target ranking code into the same decoding point cloud block.

The decoding points in the point cloud to be decoded are ranked from small to large according to a ranking code, and the ranking code includes either a Hilbert code or a Morton code. In other words, the decoding point in the point cloud to be decoded is a Hilbert code. The codes are ranked from smallest to largest, or the decoding points in the point cloud to be decoded are ranked according to the overall code from smallest to largest.

Below, the method will be explained using an example in which the decoding points in the point cloud to be decoded are ranked according to the Morton code from smallest to largest.

After obtaining the second target shift value, the Hilbert code corresponding to each decoding point is shifted to the right using the second target shift value to obtain a new Hilbert code corresponding to each decoding point, and the new Hilbert code is It is determined by the target Hilbert code, and the target Hilbert code can be understood as a second target ranking code.

The second target shift value represents the number of bits shifted in the process of shifting the decoding point, and the second target shift value may also represent the density of the point cloud to be decoded. The smaller the target shift value, the greater the number of bits in the point cloud to be decoded. It must be understood that the density of each decoding point is high.

Furthermore, the cloud of coding points to be decoded is divided according to the target Hilbert code corresponding to each decoding point, and the decoding points with the same target Hilbert code are divided into the same decoding point cloud block, that is, among one decoding point cloud block. The target Hilbert code corresponding to each decoding point is the same.

In this embodiment, a shift is performed on the ranking code of each decoding point in the point cloud to be decoded, and the point cloud to be decoded is divided into decoding point cloud blocks according to the second target ranking code corresponding to each shifted decoding point. . Here, the fact that the second target ranking code of the decoding points belonging to the same decoding point cloud block is the same indicates that the properties of these decoding points have spatial correlation, and the decoding points for which the property information has spatial correlation are assigned to the same decoding point cloud block. Divide.

Optionally, before performing a shift on the ranking code corresponding to each decoding point using the second target shift value to obtain a second target ranking code corresponding to each decoding point, the method includes:

determining the second target shift value based on a first shift parameter and a second shift parameter; or

determining the second target shift value based on a geometric quantification step corresponding to the point cloud to be decoded; or

determining a preset shift value as the second target shift value; or

and obtaining the second target shift value from the code stream to be decoded.

In this embodiment, coordinate conversion is performed on the geometric information of the point cloud to be decoded so that the point cloud is all contained within the bounding box. Here, a three-dimensional rectangular coordinate system can be formed centered on the point cloud to be decoded, the bounding box can be understood as a hexahedron covering the point cloud to be decoded, and each corner of the hexahedron is located on the X and Y axes of the coordinate system. Or it is located on the Z axis. A first shift parameter corresponding to the point cloud to be decoded is determined based on the size of the corresponding bounding box, that is, the side lengths of the corresponding bounding box in the X-axis, Y-axis, and Z-axis of the coordinate system.

Obtain the quantity of decoding points included in the point cloud to be decoded, and determine a second shift parameter corresponding to the point cloud to be decoded based on the quantity of decoding points included in the decoding point cloud.

The embodiment also has a shift calculation formula, a first shift parameter calculation formula, and a second shift parameter calculation formula set in advance. The first shift parameter and the second shift parameter are input into a calculation formula to obtain a second target shift value corresponding to the point cloud to be decoded. Here, the first shift parameter can be obtained by inputting the side length corresponding to the bounding box into the first shift parameter calculation formula; The second shift parameter can be obtained by inputting the quantity of decoding points included in the point cloud to be decoded into the second shift parameter calculation formula.

Specifically, the shift calculation formula, the first shift parameter calculation formula, and the second shift parameter calculation formula are as follows.

Here, shiftBits represents the second target shift value, MaxBits represents the first shift parameter, MinBits represents the second shift parameter, x represents the side length of the corresponding bounding box on the represents the side length of the bounding box corresponding to the Y-axis, z represents the side length of the bounding box corresponding to the Z-axis of the coordinate system, and voxelCount represents the quantity of decoding points included in the point cloud to be decoded.

For example, when the first shift parameter is 9 and the second shift parameter is 3, the second target shift value is 3.

In an optional implementation method, the second target shift value is determined based on the geometric quantification step of the decoding point cloud, for example, the second target shift value is set to a value corresponding to the geometric quantification step, or The result of adding the numerical value corresponding to the geometric quantification step and the preset numerical value is set as the second target shift value.

As a selectable implementation method, the second target shift value may be set to self-definition.

In an optional implementation, the second target shift value may be read from the code stream to be decoded.

Optionally, the step of dividing the decoding point cloud block into M decoding point cloud groups based on the quantity of decoding points included in the decoding point cloud block and a preset maximum transformation order,

If the quantity of decoding points included in the decoding point cloud block is less than or equal to the preset maximum transformation order, determining the decoding point cloud block as one decoding point cloud group;

If the decoding point quantity included in the decoding point cloud block is greater than the preset maximum transformation order, based on the decoding point quantity included in the decoding point cloud block and the first preset value, the decoding point cloud block is converted to at least It involves splitting into two decoding point cloud groups.

In this embodiment, the maximum transformation order is preset, and the preset maximum transformation order is intended to indicate the maximum order corresponding to the transformation matrix used in the inverse transformation process. Obtain the quantity of coding points included in the coding point cloud block, and if the quantity of coding points is less than or equal to the preset maximum transformation order, the coding point cloud block is determined as one coding point cloud group, that is, the corresponding coding point cloud block is obtained. Point cloud grouping is not performed on coding point cloud blocks.

If the corresponding coding point quantity is greater than the preset maximum transformation order, based on the corresponding coding point quantity and the first preset value, point cloud grouping is performed on the coding point cloud block, and the coding point cloud block is divided into at least two Divide the coding point cloud into groups. Here, based on the quantity of coding points included in the coding point cloud block and the first preset value, the specific method of performing point cloud grouping for the coding point cloud block may refer to subsequent embodiments.

Optionally, the step of dividing the decoding point cloud block into at least two decoding point cloud groups based on the decoding point quantity included in the decoding point cloud block and a first preset value, comprising:

If the first value is greater than the preset maximum transformation order, performing a grouping calculation operation on the first value using the first preset value to obtain a second value;

According to the preset decoding order of decoding points among the decoding point cloud blocks, at least some of the decoding points among the decoding points that have not performed point cloud grouping are selected until all decoding points of all decoding point cloud blocks complete point cloud grouping. It includes dividing into one decoding point cloud group.

The first numerical value represents the decoding point quantity without performing point cloud grouping of the decoding point cloud block, and when the first numerical value is greater than the preset maximum transformation order, grouping for the first numerical value is performed using the first preset numerical value. A calculation operation is performed to obtain a second value, where the second value is smaller than a preset maximum transformation order. Here, the specific method of how to use the first preset numerical value to perform a grouping calculation operation on the first numerical value to obtain the second numerical value may refer to subsequent embodiments.

After obtaining the second numerical value, at least some decoding points among decoding points for which point cloud grouping has not been performed are divided into one decoding point cloud group according to a preset decoding order of decoding points among decoding point cloud blocks. In other words, determine some decoding points among decoding point cloud blocks that have not performed point cloud grouping, and split these some decoding points into one decoding point cloud group, or all decoding point cloud blocks that have not performed point cloud grouping. Decoding points are determined, and these decoding points are divided into one decoding point cloud group. Here, the quantity of decoding points divided into the at least one decoding point cloud group is equal to the second value.

After performing point cloud grouping for some decoding points, update the first value, and if the updated first value is still greater than the preset maximum transformation order, all decoding points in the decoding point cloud block complete the point cloud grouping. Until this happens, point cloud grouping is continuously performed on at least some decoding points among the decoding point cloud blocks.

Optionally, the step of performing a grouping calculation operation on the first numerical value using the first preset numerical value to obtain a second numerical value, comprising:

If the first numerical value is greater than the preset maximum conversion order, rounding a result of dividing the first numerical value and the first preset numerical value to obtain a target numerical value;

Updating the first value to the target value until the target value is less than or equal to a preset maximum conversion order, and determining the second value to be the target value.

The process of the grouping calculation operation is as follows: performing a division operation on a third numerical value and a first preset numerical value to obtain a division result; The target value is obtained by rounding the division result. Here, the division result can be rounded up or down or rounded off. When the target value is greater than the preset maximum conversion order, update the first value to the target value, and continue for the first value and the first preset value until the target value is less than or equal to the preset maximum conversion order. Perform division association to determine the target number as the second number.

To explain in detail the technical method of performing point cloud grouping on decoding point cloud blocks, the example below will be used.

Determining the size relationship between Mj1 and Kmax, Mj1 represents the decoding point quantity without performing point cloud grouping of the decoding point cloud block, that is, the first value, and Kmax is equal to the preset maximum transformation order.

If Mj1 is greater than Kj, a grouping calculation operation is performed on the first value . That is, the division result of the first value and 2 is rounded down to obtain the target value, where Mj in the formula represents the target value, and the first preset value is 2. It should be understood that in other embodiments, the first preset value may be set to self-definition.

If the target value is greater than Kmax, the first value is updated to the target value, and the grouping calculation operation is repeated until the first value is less than or equal to the preset maximum transformation order. In this case, the last updated Mj1 is determined as the second value, and some decoding points are divided into one decoding point cloud group according to the preset decoding order of each decoding point in the decoding block, where the partial decoding points The quantity is the same as the second number.

Then, the size relationship between Mj2 and Kj is determined, where Mj2 represents the quantity of decoding points for which point cloud grouping of the decoding point cloud block has not been performed, and in the above step, grouping into point clouds has already been performed for some decoding points. Since it was performed, the value of Mj2 is the difference between the quantity of all decoding points among the decoding point cloud blocks and the last updated Mj1. In other words, the value of Mj2 is smaller than the value of Mj1. If Mj2 is greater than Kj, a grouping calculation operation is performed on Mj2 and point cloud grouping is performed on the decoding block according to the calculation result. A specific implementation method is to perform grouping into a point cloud on the decoding block according to Mj1. It is consistent with the above implementation method, and will not be repeated here.

If the quantity of decoding points for which point cloud grouping of the decoding point cloud block has not been performed is less than or equal to Kj, performing point cloud grouping on the decoding block is stopped.

Furthermore, you may refer to the examples below.

The decoding point cloud block includes 12 decoding points without performing point cloud grouping, the maximum transformation order is preset to 5, and the first preset number is 2.

In this case, a grouping calculation operation is performed on the first numerical value 12 and the first preset numerical value 2 to obtain the target numerical value 6. Additionally, if the target value 6 is greater than the preset maximum transformation order 5, the first value is updated to the target value, that is, the first value is updated to 6. Additionally, a grouping calculation operation is performed on the updated first number 6 to obtain the target number 3, and at this time, if the target number 3 is less than the preset maximum transformation order 5, the preset decoding order of each decoding point in the decoding point cloud block Accordingly, three decoding points are selected and divided into the same decoding point cloud group.

After performing primary point cloud grouping on the decoding point cloud block, update the first numerical value to 9, and perform grouping calculation operation on the first numerical value 9 and the first preset numerical value 2 to obtain the target numerical value 4. . In addition, if the target number 4 is less than the preset maximum transformation order 5, 4 decoding points are selected among the decoding point cloud blocks according to the preset decoding order of each decoding point and divided into the same decoding point cloud group.

After performing secondary point cloud grouping on the decoded point cloud block, the first numerical value is updated to 5. In addition, if the first numerical value 5 is equal to the preset maximum transformation order 5, five decoding points are selected among the decoding point cloud blocks according to the preset decoding order of each decoding point and divided into the same decoding point cloud group.

In summary, for a decoding point cloud block containing 12 decoding points without performing point cloud grouping, the corresponding decoding point cloud block is divided into three point cloud groups, and the decoding points corresponding to these three point cloud groups are The quantities can be 3, 4, or 5, respectively.

Optionally, before performing inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain the target high-frequency coefficient and the target low-frequency coefficient, the method includes:

Obtaining a second transformation ID in the point cloud to be decoded;

If the second transformation ID indicates that inverse transformation is to be performed on all decoding point cloud groups, determining all decoding point cloud groups as the third target decoding point cloud group;

When the second transformation ID indicates that self-adaptive inverse transformation is performed on all decoding point cloud groups, determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group Includes steps.

It should be understood that the third target decoding point cloud grouping refers to a point cloud grouping for which inverse transformation is performed, that is, inverse transformation must be performed on each decoding point in the third target decoding point cloud group.

In this embodiment, the point cloud to be decoded has a second transformation ID preset, and the second transformation ID is also called a self-adaptive transformation ID; optionally, the second transformation ID and the first transformation ID are the same transformation ID. It should be understood that the second transformation ID can be expressed as AdpTransform, and the second transformation ID can be set as self-defined.

If AdpTransform=0, it indicates that inverse transformation is performed on all decoding points. In this case, all decoding point cloud groups are determined as the third target decoding point cloud group.

If AdpTransform=1, it indicates that each decoding point cloud group performs its own adaptive inverse transformation, and if so, it is determined whether to perform inverse transformation for that decoding point cloud group according to the actual situation of each decoding point cloud group, in this case , At least one decoding point cloud group among the N decoding point cloud groups is determined as the target decoding point cloud group.

Optionally, the step of determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group,

Obtaining a third conversion ID corresponding to the first point cloud group;

When the third transformation ID indicates that inverse transformation is to be performed on the first point cloud group, determining the point cloud group as the third target decoding point cloud group.

It should be understood that each decoding point cloud group among the decoding point cloud blocks is ranked according to the grouping time order of the point cloud group, and the decoding point cloud group ranked most forward among the decoding point cloud blocks is determined as the first point cloud group.

In this embodiment, the third transformation ID is preset in the decoding point cloud block, and the third transformation ID is used to transform the first point cloud group into the third target decoding point cloud when inverse transformation is performed on the first point cloud group. It is meant to represent a decision made as a group. In other words, a third transformation ID is preset in the decoding point cloud block, and it is determined whether the first point cloud group is the third target decoding point cloud group according to the third transformation ID.

Optionally, the step of determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group,

For a second point cloud group, determining a first reconstruction value and a second reconstruction value corresponding to the second point cloud group;

calculating an absolute value of the difference between the first reconstruction value and the second reconstruction value;

When the absolute value is smaller than a second preset value, determining the second point cloud group as a third target decoding point cloud group.

Each decoding point cloud group among the decoding point cloud blocks is ranked according to the grouping time order of the point cloud group, and the decoding point cloud group excluding the most forward-ranked decoding point cloud group among the decoding point cloud blocks is determined as the second point cloud group. You must understand what you are doing. For example, a decoding point cloud block includes three decoding point cloud groups, and the ranked second and third ranked decoding point cloud groups are determined as the second point cloud group.

For a second point cloud group, a decoding point cloud group adjacent to the second point cloud group among decoding point cloud blocks and located in front of the second point cloud group is determined as the adjacent point cloud group, and the adjacent point cloud group is determined. The maximum value among the reconstruction information corresponding to the cloud group is determined as the first reconstruction value, and the minimum value among the reconstruction information corresponding to the adjacent point cloud group is determined as the second reconstruction value.

The reconstruction information corresponding to the adjacent point cloud group is a set of reconstruction information corresponding to each decoding point among the adjacent point cloud group. In other words, the first reconstruction value is provided to each decoding point among the adjacent point cloud group. It should be understood that it is the maximum value among the corresponding reconstruction information, and the second reconstruction value is the minimum value among the reconstruction information corresponding to each decoding point among the adjacent point cloud groups.

A difference value result between the first reconstruction value and the second reconstruction value is calculated, and the absolute value of the difference value result is obtained. Furthermore, in this embodiment, a second preset value is also set, and when the absolute value is smaller than the second preset value, the second point cloud group is determined as the third target coding point cloud group.

The third transform ID may be expressed as a transform ID (TransformFlag), and if TransformFlag=1, the corresponding decoding point cloud group is determined as the third target decoding point cloud group; It should be understood that if TransformFlag=0, the corresponding decoding point cloud group is not determined as the third target decoding point cloud group.

To facilitate understanding, the following formula can be introduced to further explain the technical solution of this embodiment.

In the case of TransformFlag = 0, the decoding point cloud group is not determined as the third target decoding point cloud group.

Here, AttRecmax represents the first reconstruction value, AttRecmin represents the second reconstruction value, and Threshold represents the second preset value.

In the case of, TransformFlag=1 is set, and the decoding point cloud group is determined as the third target decoding point cloud group.

Optionally, the step of performing inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain the target high-frequency coefficient and the target low-frequency coefficient:

Inverse quantification is performed on the first high frequency coefficient and the first low frequency coefficient using a preset first quantification step and a preset second quantification step, respectively, so that the inverse quantified first high frequency coefficient and the inverse quantified first low frequency obtaining coefficients;

performing inverse quantification on the second high-frequency coefficient and the second low-frequency coefficient using a preset third quantification step to obtain an inverse-quantified second high-frequency coefficient and an inverse-quantified second low-frequency coefficient;

determining the inversely quantified first high frequency coefficient and the inversely quantified second high frequency coefficient as the target high frequency coefficient;

and determining the inversely quantified first low-frequency coefficient and the inversely quantified second low-frequency coefficient as the target low-frequency coefficient.

The color property of the second conversion coefficient is the YUV color space, and the U and V components in the YUV color space are determined as the first component, the Y component is determined as the second component, and the high frequency coefficient of the first component is determined as the first component. The high-frequency coefficient may be determined as the high-frequency coefficient, the low-frequency coefficient of the first component may be determined as the first low-frequency coefficient, the high-frequency coefficient of the second component may be determined as the second high-frequency coefficient, and the high-frequency coefficient of the second component may be determined as the second high-frequency coefficient. there is.

Inverse quantification is performed on the first high-frequency coefficient using a first quantification step preset for the first high-frequency coefficient, to obtain an inverse-quantified first high-frequency coefficient. Here, the first quantification step is equal to the sum of the initial step, conversion step, and high frequency step, and the initial step, conversion step, and high frequency step can be set to self-definition.

It should be understood that the first quantification step can be determined by the formula below.

Here, Qfin1 represents the first quantification step, Qori represents the initial step, Offsetcoeff represents the conversion step, and OffsetAC represents the high frequency step.

Inverse quantification is performed on the first low-frequency coefficient using a second quantification step preset for the first low-frequency coefficient, to obtain the inverse-quantified first low-frequency coefficient. Here, the second quantification step is equal to the sum of the initial step, conversion step, and low-frequency step, and the initial step, conversion step, and low-frequency step can be set to self-definition.

It should be understood that the second quantification step can be determined by the formula below.

Here, Qfin2 represents the second quantification step, and OffsetDC represents the low frequency step.

For the second high-frequency coefficient and the second low-frequency coefficient, inverse quantification is performed on the second high-frequency coefficient and the second low-frequency coefficient using a preset third quantification step, such that the inverse-quantified second high-frequency coefficient and the inverse-quantified first 2 Obtain the low-frequency coefficient. wherein the third quantification step is equal to the sum of the initial step, the conversion step and the low frequency step, or the third quantification step is equal to the sum of the initial step, the conversion step and the high frequency step, and the initial step, the conversion step, the high frequency step and the low frequency step are equal to the sum of the initial step, the conversion step and the low frequency step. The steps can all be set to self-definition, and the high-frequency step and low-frequency step are the same.

It should be understood that the third quantification step can be determined by the formula below.

Here, Qfin3 represents the third quantification step, and OffsetDC in the above formula can be replaced with OffsetAC.

In this embodiment, the inversely quantified first high-frequency coefficient and the inversely quantified second high-frequency coefficient are determined as the target high-frequency coefficient; The inversely quantified first low-frequency coefficient and the inversely quantified second low-frequency coefficient are determined as the target low-frequency coefficient.

It should be explained that the subject of execution of the point cloud attribute coding method provided by the embodiment of the present application may be a coder or a control module that performs the point cloud attribute coding method among the coders. Taking the coder performing the point cloud attribute coding method in the embodiment of the present application as an example, the coder provided by the embodiment of the present application will be described.

As shown in Figure 6, the coder 300,

A first acquisition module 301 that acquires a point cloud to be coded;

Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. first division module 302;

For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. a second division module 303 for division;

a first coding module 304 that performs transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient;

and a second coding module 305 that generates a target code stream by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group.

Optionally, the first segmentation module 302 specifically:

Using the first target shift value, perform a shift on the ranking code corresponding to each coding point to obtain the first target ranking code corresponding to each coding point;

Coding points with the same first target ranking code are divided into the same coding point cloud block.

Optionally, the second segmentation module 303 specifically:

If the quantity of coding points included in the coding point cloud block is less than or equal to the preset maximum transformation order, determine the coding point cloud block as one coding point cloud group;

When the quantity of coding points included in the coding point cloud block is greater than the preset maximum transformation order, based on the quantity of coding points included in the coding point cloud block and the first preset value, the coding point cloud block is converted to at least Split into two coding point cloud groups.

Optionally, the coder 300:

a second acquisition module that acquires a first transformation ID in the point cloud to be coded;

When the first transformation ID indicates that transformation coding is performed on all coding point cloud groups, a first determination module that determines all coding point cloud groups as the first target coding point cloud group;

When the first transformation ID indicates that self-adaptive transformation coding is performed on all coding point cloud groups, at least one coding point cloud group among the N coding point cloud groups is determined as the first target coding point cloud group. It further includes a second decision module.

Optionally, the first coding module 304 specifically:

calculate a multiplication result between a transformation matrix and the first residual information, and determine the multiplication result as a first transformation coefficient to be quantified;

Quantification is performed on the first conversion coefficient to be quantified to obtain the quantified first conversion coefficient.

The coder provided by the embodiment of the present application can implement each process implemented by the method embodiment shown in FIG. 4 and also achieves the same technical effect, so detailed description is omitted here to prevent duplication.

It should be explained that the subject of execution of the point cloud attribute decoding method provided by the embodiment of the present application may be a decoder or a control module that performs the point cloud attribute decoding method in the decoder. The decoder provided by the embodiment of the present application will be described taking as an example that the decoder performs the point cloud attribute decoding method in the embodiment of the present application.

As shown in Figure 7, the decoder 400,

A decoding module 401 that performs decoding on a code stream to be decoded to obtain a point cloud to be decoded;

Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. third division module 402;

For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the decoding point cloud block and the preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. a fourth division module 403 for division;

an inverse quantification module 404 that performs inverse quantification processing on the transformation coefficients corresponding to the third target decoding point cloud group to obtain target high-frequency coefficients and target low-frequency coefficients;

and an inverse transformation module 405 that performs inverse transformation on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group.

Optionally, the third segmentation module 402 specifically:

Using the second target shift value, perform a shift on the ranking code corresponding to each decoding point to obtain a second target ranking code corresponding to each decoding point;

The second target ranking code partitions the same decoding point into the same decoding point cloud block.

Optionally, the third segmentation module 402 specifically:

determine the second target shift value based on a first shift parameter and a second shift parameter; or

determine the second target shift value based on a geometric quantification step corresponding to the point cloud to be decoded; or

determine a preset shift value as the second target shift value; or

The second target shift value is obtained from the code stream to be decoded.

Optionally, the fourth segmentation module 403 specifically:

If the quantity of decoding points included in the decoding point cloud block is less than or equal to the preset maximum transformation order, determining the decoding point cloud block as one decoding point cloud group;

If the decoding point quantity included in the decoding point cloud block is greater than the preset maximum transformation order, based on the decoding point quantity included in the decoding point cloud block and the first preset value, the decoding point cloud block is converted to at least Split into two decoding point cloud groups.

Optionally, the fourth segmentation module 403 specifically:

If the first value is greater than the preset maximum transformation order, use the first preset value to perform a grouping calculation operation on the first value to obtain a second value;

According to the preset decoding order of decoding points among the decoding point cloud blocks, at least some of the decoding points among the decoding points that have not performed point cloud grouping are selected until all decoding points of all decoding point cloud blocks complete point cloud grouping. When divided into one decoding point cloud group, the at least some decoding point quantities are equal to the second value.

Optionally, the fourth segmentation module 403 specifically:

If the first value is greater than the preset maximum conversion order, the result of dividing the first value and the first preset value is rounded to obtain a target value;

The first value is updated to the target value until the target value is less than or equal to a preset maximum conversion order, and the second value is determined as the target value.

Optionally, the decoder 400:

a third acquisition module that acquires a second transformation ID in the point cloud to be decoded;

When the second transformation ID indicates that inverse transformation is to be performed on all decoding point cloud groups, a third determination module for determining all decoding point cloud groups as the third target decoding point cloud group;

When the second transformation ID indicates that self-adaptive inverse transformation is performed on all decoding point cloud groups, determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group Includes a fourth decision module.

Optionally, the third decision module specifically:

Obtain a third transformation ID corresponding to the first point cloud group;

If the third transformation ID indicates that inverse transformation is to be performed on the first point cloud group, the point cloud group is determined as the third target decoding point cloud group.

Optionally, the fourth decision module specifically:

For a second point cloud group, determine a first reconstruction value and a second reconstruction value corresponding to the second point cloud group;

Calculate an absolute value of the difference between the first reconstruction value and the second reconstruction value;

If the absolute value is smaller than the second preset value, the second point cloud group is determined as the third target decoding point cloud group.

Optionally, the inverse quantification module 404 specifically:

Inverse quantification is performed on the first high frequency coefficient and the first low frequency coefficient using a preset first quantification step and a preset second quantification step, respectively, so that the inverse quantified first high frequency coefficient and the inverse quantified first low frequency obtain coefficients;

Perform inverse quantification on the second high-frequency coefficient and the second low-frequency coefficient using a preset third quantification step to obtain an inverse-quantified second high-frequency coefficient and an inverse-quantified second low-frequency coefficient;

determining the inversely quantified first high-frequency coefficient and the inversely quantified second high-frequency coefficient as the target high-frequency coefficient;

The inversely quantified first low-frequency coefficient and the inversely quantified second low-frequency coefficient are determined as the target low-frequency coefficient.

In an embodiment of the present application, after obtaining a point cloud to be coded, the point cloud to be coded is divided into at least one coding point cloud block, point cloud grouping is performed on the coding points in each coding point cloud block, and coding is performed. Split the point cloud block into N coded point cloud groups. In the transform coding process, the correlation between residual information corresponding to each coding point is sufficiently considered, and transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient. is obtained, where there is a correlation between the residual information of each coding point in the first target coding point cloud group. Furthermore, entropy coding is performed on the quantified first transform coefficient and the second residual information corresponding to the second target coding point cloud group to generate a target code stream. The coding process sufficiently considers the correlation between residual information corresponding to each coding point and performs transform coding on the first residual information corresponding to the first target coding point cloud group to reduce redundant information in the transform coefficient and residual information. The efficiency of point cloud attribute coding has been improved.

The coder and decoder in the embodiments of the present application may be a device, a device equipped with an operating system, or an electronic device, and may also be a component, an integrated circuit, or a chip in the device. The device or electronic device may be a mobile terminal or a non-mobile terminal. Illustratively, mobile terminals may include, but are not limited to, the types of terminals 11 listed above, and non-mobile terminals may include servers, network attached storage (NAS), and personal computers. It may be a PC), a television (TV), an automatic teller machine, or a self-service device, and the embodiment of the present application is not specifically limited.

The coder provided by the embodiment of the present application can implement each process implemented by the method embodiment shown in FIG. 4 and also achieves the same technical effect, so detailed description is omitted here to prevent duplication.

The decoder provided by the embodiment of the present application can implement each process implemented by the method embodiment shown in FIG. 5 and also achieves the same technical effect, so detailed description is omitted here to prevent duplication.

Optionally, as shown in FIG. 8, embodiments of the present application also provide an electronic device 500, comprising a processor 501, a memory 502, and stored in the memory 502, wherein the processor 501 ) and includes a program or command that can be executed on, for example, when the communication device 500 is a terminal and the program or command is executed by the processor 501, each process of the point cloud attribute coding method embodiment implements and achieves the same technical effect, or implements each process of the above-described point cloud attribute decoding method embodiments and achieves the same technical effect.

Embodiments of the present application also provide a terminal including a processor and a communication interface, where the processor includes:

Obtain a point cloud to code;

Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block; ;

For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. to divide;

Performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient;

An operation of generating a target code stream is performed by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group.

Alternatively, the processor may:

Perform decoding on the code stream to be decoded to obtain a point cloud to be decoded;

Based on the quantity of decoding points included in the point cloud to be decoded, volume information of a bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block; ;

For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the point cloud block to be decoded and a preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. Divide into;

Perform inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain target high-frequency coefficients and target low-frequency coefficients; Inverse transformation processing is performed on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group.

The terminal embodiment corresponds to the terminal-side method embodiment, and each implementation process and implementation method of the method embodiment can be applied to the terminal embodiment, and the same technical effect can be achieved. Specifically, Figure 8 is a hardware structure diagram implementing a terminal of an embodiment of the present application.

The terminal 1000 includes a radio frequency unit 1001, a network module 1002, an audio output unit 1003, an input unit 1004, a sensor 1005, a display unit 1006, a user input unit 1007, It includes, but is not limited to, components such as the interface unit 1008, memory 1009, and processor 1010.

Technicians in the art know that the terminal 1000 may also include a power source (e.g., a battery) that supplies power for each component, and the power source is connected to the processor 1010 through a power management system. Through this, you will understand that functions such as charging, discharge management, and power consumption management can be implemented. It will be understood that the terminal structure shown in FIG. 9 does not limit the terminal, and that the terminal may include more or fewer components than those shown, or may be constructed by combining some components or different components. Detailed description will be omitted.

In an embodiment of the present application, the input unit 1004 may include a graphics processing unit (GPU) 10041 and a microphone 10042, and the image processing unit 10041 may be used in a video capture mode or image capture. It will be understood that in this mode, an image capture device (e.g., a camera) performs processing on image data, either a static image or a video, acquired. The display unit 1006 may include a display panel 10061, and optionally, the display panel 10061 may be configured using a liquid crystal display, an organic light emitting diode, or the like. The user input unit 1007 includes a touch panel 10071 and other input devices 10072. The touch panel 10071 may also be referred to as a touch screen. The touch panel 10071 may include two parts: a touch detection device and a touch controller. Other input devices 10072 may include one or more of a physical keyboard, function keys (e.g., volume control buttons, switch buttons, etc.), a trackball, a mouse, a stick, etc., but are not limited thereto.

In the embodiment of the present application, the radio frequency unit 1001 receives downlink data from the network-side device and then transmits it to the processor 1010 for processing; Then, uplink data is transmitted to the network-side device. Typically, the radio frequency unit 1001 may include an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier, a duplexer, etc., but is not limited thereto.

The memory 1009 can store software programs or instructions and various data. Memory 1009 may primarily include a program or instruction storage area and a data storage area, wherein the program or command storage area contains an operating system, an application program or command required for at least one function (e.g., a sound playback area). function, image play function, etc.) can be saved. In addition, the memory 1009 may include high-speed random access memory, and may also include non-volatile memory, where the non-volatile memory includes read-only memory (ROM), programmable memory (Programmable Memory), and non-volatile memory. It may be ROM, PROM), volatile programmable memory (Erasable PROM, EPROM), electrically volatile programmable memory (EPROM, EEPROM), or flash. For example, at least one disk storage device, flash device, or other non-volatile solid-state storage device.

Processor 1010 may include one or multiple processing units; Optionally, the processor 1010 may be integrated with an application processor and a modulation/demodulation processor, where the application processor mainly processes the operating system, user interface and applications or instructions, and the modulation/demodulation processor mainly processes the operating system, user interface, and applications or commands. It handles wireless communications, for example a baseband processor. It will be appreciated that the modulation/demodulation processor may also not be integrated into processor 1010.

Here, the processor:

Obtain a point cloud to code;

Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block; ;

For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into M coding point cloud groups. to divide;

Performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient;

An operation of generating a target code stream is performed by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group.

Alternatively, the processor may:

Perform decoding on the code stream to be decoded to obtain a point cloud to be decoded;

Based on the quantity of decoding points included in the point cloud to be decoded, volume information of a bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block; ;

For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the point cloud block to be decoded and a preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. Divide into;

Perform inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain target high-frequency coefficients and target low-frequency coefficients; Inverse transformation processing is performed on the target high-frequency coefficient and the target low-frequency coefficient to obtain residual information corresponding to the third target decoding point cloud group.

An embodiment of the present invention also provides a readable storage medium, the readable storage medium may be non-volatile or volatile, the readable storage medium stores a program or command, and the program or when an instruction is executed by a processor, it implements each process of the point cloud attribute coding method embodiment, or implements each process of the point cloud attribute decoding method embodiment, and also achieves the same technical effect, preventing duplication. For this reason, detailed description will be omitted here.

An embodiment of the present invention also provides a computer program product, wherein the computer program product is stored in a non-transitory readable storage medium, and the computer program product is executed by at least one processor to perform the point cloud attribute coding method. Since each process of the embodiment is implemented or each process of the point cloud attribute decoding method embodiment is implemented and has the same technical effect, detailed description is omitted here to prevent duplication.

Here, the processor is a processor in the terminal in the above embodiment. The readable medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

An embodiment of the present application also provides a chip, wherein the chip includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor executes a program or instruction to code the point cloud attribute. Since each process of the method embodiment is implemented, or each process of the point cloud attribute decoding method embodiment is implemented, and also produces the same technical effect, detailed description is omitted here to prevent duplication.

It will be understood that the chips referred to in the embodiments of the present application may also be referred to as system level chips, system chips, chips of system-on-a-chip or system-on-a-chip, etc.

It should be noted that, in the text, the term “comprise” or any other variant thereof means non-exclusively including, so that a process, method, article or device that includes a set of elements only includes those elements. In addition, it may also include other elements not explicitly mentioned, or elements unique to such process, method, product or device. Without further limitation, any element defined by the phrase "comprising a..." excludes the inclusion of any other identical element in any process, method, article or device incorporating that element. I never do that. In addition, it should be pointed out that the scope of the methods and devices in the implementation mode of the present application is not limited to functions executed according to the order exemplified or discussed, and also according to the mentioned functions in an essentially simultaneous manner or It may include functions that are executed in the opposite order, for example, the above-described method may be executed differently from the described order, and various steps may be added, omitted, or combined. Also, the features described in some examples can be referred to and combined in other examples.

Through the description of the above implementation method, those skilled in the art can see that the method of the above embodiment can be implemented by adding a general-purpose hardware platform required for software, and of course, it can be implemented through hardware. Among many cases, the former is a more preferable implementation method. Based on this, the essence of the technical solution of this application or the part that contributes to the prior art can be implemented in the form of a computer software product, and the computer software product may be implemented in the form of a storage medium (e.g. ROM/RAM, magnetic disk) , optical disk), and some instructions are included to enable a terminal (which may be a mobile phone, computer, server, air conditioner, or network device) to implement the method of each embodiment of the present application.

Although the embodiments of the present application have been described above in connection with the drawings, the present application is not limited to the above specific embodiments, and the above specific embodiments are only illustrative and not restrictive, and those skilled in the art will understand the present application. Under the hint, as long as the idea and claims of the present application do not go beyond the scope of protection, many more forms can be implemented, and all of them will fall within the scope of protection of the present application.

Claims (22)

In the point cloud attribute coding method,
Obtaining a point cloud to be coded, ranking coding points in the point cloud to be coded according to a preset coding order;
Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. step;
For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. dividing, where N is an integer;
Transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a first quantified transform coefficient, and the first target coding point cloud group is the N coding point clouds. at least part of a group, wherein the first target coding point cloud group is a coding point cloud group that performs transform coding;
A target code stream is generated by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group, wherein the second target coding point cloud group is the N coding A point cloud attribute coding method comprising the step of being a coding point cloud group excluding the first target coding point cloud group among the point cloud groups. According to paragraph 1,
Coding points in the point cloud to be coded are ranked from small to large according to a ranking code, and the ranking code includes one of a Hilbert code and a Morton code;
Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. The above steps are:
Using the first target shift value, the ranking code corresponding to each coding point is shifted to obtain the first target ranking code corresponding to each coding point, and the target shift value is stored in the point cloud to be coded. Determining based on the included coding point quantity and volume information of the bounding box;
A point cloud attribute coding method comprising dividing coding points having the same first target ranking code into identical coding point cloud blocks. According to paragraph 1,
The step of dividing the coding point cloud block into N coding point cloud groups based on the quantity of coding points included in the coding point cloud block and a preset maximum transformation order,
If the quantity of coding points included in the coding point cloud block is less than or equal to the preset maximum transformation order, determining the coding point cloud block as one coding point cloud group;
When the quantity of coding points included in the coding point cloud block is greater than the preset maximum transformation order, based on the quantity of coding points included in the coding point cloud block and the first preset value, the coding point cloud block is converted to at least A point cloud attribute coding method that includes splitting into two coding point cloud groups. According to paragraph 1,
Before performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient, the method includes:
Obtaining a first transformation ID in the point cloud to be coded;
When the first transformation ID indicates that transformation coding is performed on all coding point cloud groups, determining all coding point cloud groups as the first target coding point cloud group;
When the first transformation ID indicates that self-adaptive transformation coding is performed on all coding point cloud groups, at least one coding point cloud group among the N coding point cloud groups is determined as the first target coding point cloud group. Point cloud attribute coding method including the steps of: According to paragraph 1,
The step of performing transform coding and quantification processing on the first residual information corresponding to the first target coding point cloud group to obtain a quantified first transform coefficient,
A multiplication result between a transformation matrix and the first residual information is calculated, and the multiplication result is determined as a first transformation coefficient to quantify, where the order of the first transformation matrix is the coding included in the first target coding point cloud group. Steps determined based on the point quantity;
A point cloud attribute coding method comprising performing quantification on the first transform coefficient to be quantified to obtain the quantified first transform coefficient. In the point cloud attribute decoding method,
Performing decoding on a code stream to be decoded to obtain a point cloud to be decoded, wherein decoding points in the point cloud to be decoded are ranked according to a preset coding order;
Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. step;
For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the decoding point cloud block and the preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. dividing, where M is an integer;
Inverse quantification processing is performed on the second transform coefficient corresponding to the third target decoding point cloud group to obtain target high-frequency coefficients and target low-frequency coefficients, wherein the third target decoding point cloud group is the M decoding point cloud group. wherein the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing;
A point cloud attribute decoding method comprising performing inverse transformation processing on the target high-frequency coefficient and the target low-frequency coefficient to obtain third residual information corresponding to the third target decoding point cloud group. According to clause 6,
The decoding points in the point cloud to be decoded are ranked from small to large according to a ranking code, and the ranking code includes one of a Hilbert code and a Morton code;
Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. The above steps are:
Using the second target shift value, a ranking code corresponding to each decoding point is shifted to obtain a second target ranking code corresponding to each decoding point, where the second target shift value is the point to be decoded. Determining based on the quantity of decoding points included in the cloud and volume information of the bounding box;
A point cloud attribute decoding method comprising partitioning decoding points having the same second target ranking code into the same decoding point cloud block. In clause 7,
Before obtaining a second target ranking code corresponding to each decoding point by performing a shift on the ranking code corresponding to each decoding point using the second target shift value, the method includes:
The second target shift value is determined based on a first shift parameter and a second shift parameter, where the first shift parameter is determined based on volume information of the bounding box, and the second shift parameter is the decoding method. Determining based on the quantity of decoding points included in the point cloud; or
determining the second target shift value based on a geometric quantification step corresponding to the point cloud to be decoded; or
determining a preset shift value as the second target shift value; or
Point cloud attribute decoding method comprising obtaining the second target shift value from the code stream to be decoded. According to clause 6,
The step of dividing the decoding point cloud block into M decoding point cloud groups based on the quantity of decoding points included in the decoding point cloud block and a preset maximum transformation order,
If the quantity of decoding points included in the decoding point cloud block is less than or equal to the preset maximum transformation order, determining the decoding point cloud block as one decoding point cloud group;
If the decoding point quantity included in the decoding point cloud block is greater than the preset maximum transformation order, based on the decoding point quantity included in the decoding point cloud block and the first preset value, the decoding point cloud block is converted to at least A point cloud attribute decoding method comprising splitting into two decoding point cloud groups. According to clause 9,
The step of dividing the decoding point cloud block into at least two decoding point cloud groups based on the decoding point quantity included in the decoding point cloud block and a first preset value,
When the first value is greater than the preset maximum transformation order, a grouping calculation operation is performed on the first value using the first preset value to obtain a second value, where the first value is the decoding point. Indicating a quantity of decoding points without performing point cloud grouping of cloud blocks, wherein the second number is less than or equal to a preset maximum transformation order;
According to the preset decoding order of decoding points among the decoding point cloud blocks, at least some of the decoding points among the decoding points that have not performed point cloud grouping are selected until all decoding points of all decoding point cloud blocks complete point cloud grouping. Dividing into one decoding point cloud group, wherein the at least some decoding point quantities are equal to the second numerical value. According to clause 10,
The step of performing a grouping calculation operation on the first numerical value using the first preset numerical value to obtain a second numerical value,
If the first numerical value is greater than the preset maximum conversion order, rounding a result of dividing the first numerical value and the first preset numerical value to obtain a target numerical value;
A point cloud attribute decoding method comprising updating the first value to the target value until the target value is less than or equal to a preset maximum transformation order, and determining the second value to be the target value. According to clause 6,
Before performing inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain the target high-frequency coefficient and the target low-frequency coefficient, the method includes:
Obtaining a second transformation ID in the point cloud to be decoded;
If the second transformation ID indicates that inverse transformation is to be performed on all decoding point cloud groups, determining all decoding point cloud groups as the third target decoding point cloud group;
When the second transformation ID indicates that self-adaptive inverse transformation is performed on all decoding point cloud groups, determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group Point cloud attribute decoding method including steps. According to clause 12,
The M decoding point cloud groups are ranked according to a preset order in the decoding point cloud block, the M decoding point cloud groups include one first point cloud group, and the first point cloud group is the M Among the decoding point cloud groups, it is the most forward-ranked decoding point cloud group;
The step of determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group,
Obtaining a third conversion ID corresponding to the first point cloud group;
When the third transformation ID indicates that inverse transformation is to be performed on the first point cloud group, determining the point cloud group as the third target decoding point cloud group. According to clause 12,
The M decoding point cloud groups are ranked according to a preset order in the decoding point cloud block, the M decoding point cloud groups include M - 1 second point cloud groups, and the second point cloud group is A decoding point cloud group excluding the most forward-ranked decoding point cloud group among the M decoding point cloud groups;
The step of determining at least one decoding point cloud group among the M decoding point cloud groups as the third target decoding point cloud group,
For a second point cloud group, a first reconstruction value and a second reconstruction value corresponding to the second point cloud group are determined, and the first reconstruction value is included in the reconstruction information corresponding to the adjacent point cloud group. is the maximum value, and the second reconstruction value is the minimum value among reconstruction information corresponding to an adjacent point cloud group, and the adjacent point cloud group is adjacent to the second point cloud group among the decoding point cloud blocks and the second point A decoding point cloud group located in front of the cloud group;
calculating an absolute value of the difference between the first reconstruction value and the second reconstruction value;
When the absolute value is smaller than a second preset value, determining the second point cloud group as a third target decoding point cloud group. According to clause 6,
The second conversion coefficient includes a first component and a second component, the first component includes a first high-frequency coefficient and a first low-frequency coefficient, and the second component includes a second high-frequency coefficient and a second low-frequency coefficient. Contains;
The step of performing inverse quantification processing on the second transform coefficient corresponding to the third target decoding point cloud group to obtain the target high-frequency coefficient and the target low-frequency coefficient,
Inverse quantification is performed on the first high frequency coefficient and the first low frequency coefficient using a preset first quantification step and a preset second quantification step, respectively, so that the inverse quantified first high frequency coefficient and the inverse quantified first low frequency obtaining coefficients;
performing inverse quantification on the second high-frequency coefficient and the second low-frequency coefficient using a preset third quantification step to obtain an inverse-quantified second high-frequency coefficient and an inverse-quantified second low-frequency coefficient;
determining the inversely quantified first high frequency coefficient and the inversely quantified second high frequency coefficient as the target high frequency coefficient;
A point cloud attribute decoding method comprising determining the inverse-quantified first low-frequency coefficient and the inverse-quantified second low-frequency coefficient as the target low-frequency coefficient. For coders,
A first acquisition module that acquires a point cloud to be coded, where coding points in the point cloud to be coded are ranked according to a preset coding order;
Based on the quantity of coding points included in the point cloud to be coded, the volume information of the bounding box corresponding to the point cloud to be coded, and the order of each coding point, dividing the point cloud to be coded into at least one coding point cloud block. first division module;
For each coding point cloud block containing at least two coding points, based on the quantity of coding points included in the coding point cloud block and the preset maximum transformation order, the coding point cloud block is divided into N coding point cloud groups. A second division module where N is an integer for division;
Transform coding and quantification processing are performed on the first residual information corresponding to the first target coding point cloud group to obtain a first quantified transform coefficient, and the first target coding point cloud group is the N coding point clouds. a first coding module that is at least part of a group, and wherein the first target coding point cloud group is a coding point cloud group that performs transform coding;
A target code stream is generated by performing entropy coding on the quantified first transform coefficient and second residual information corresponding to the second target coding point cloud group, wherein the second target coding point cloud group is the N coding A coder including a second coding module that is a coding point cloud group excluding the first target coding point cloud group among the point cloud groups. In the decoder,
a decoding module that performs decoding on a code stream to be decoded to obtain a point cloud to be decoded, wherein decoding points in the point cloud to be decoded are ranked according to a preset coding order;
Based on the quantity of decoding points included in the point cloud to be decoded, the volume information of the bounding box corresponding to the point cloud to be decoded, and the order of each decoding point, dividing the point cloud to be decoded into at least one decoding point cloud block. third split module;
For each decoding point cloud block containing at least two decoding points, based on the quantity of decoding points included in the decoding point cloud block and the preset maximum transformation order, the decoding point cloud block is divided into M decoding point cloud groups. A fourth division module where M is a constant number for division;
Inverse quantification processing is performed on the second transform coefficient corresponding to the third target decoding point cloud group to obtain a target high-frequency coefficient and a target low-frequency coefficient, wherein the target decoding point cloud group is at least part of the decoding point cloud group. , the third target decoding point cloud group is at least a part of the M decoding point cloud groups, and the third target decoding point cloud group is a decoding point cloud group that performs inverse quantification processing;
A decoder comprising an inverse transformation module that performs inverse transformation on the target high-frequency coefficient and the target low-frequency coefficient to obtain third residual information corresponding to the third target decoding point cloud group. A terminal comprising a memory, a processor, and a program or instruction stored in the memory and executable on the processor, and when the program or instruction is executed by the processor, the device according to any one of claims 1 to 5 A terminal implementing the steps of the point cloud attribute coding method, or implementing the steps of the point cloud attribute decoding method of any one of claims 6 to 15. In a readable storage medium, a program or instruction is stored in the readable storage medium, and when the program or instruction is executed by a processor, the step of the point cloud attribute coding method of any one of claims 1 to 5 is performed. A readable storage medium implementing or implementing the steps of the point cloud attribute decoding method of any one of claims 6 to 15. In the chip, it includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor executes a program or instruction to perform the point cloud attribute coding method of any one of claims 1 to 5. A chip implementing the steps, or implementing the steps of the point cloud attribute decoding method of any one of claims 6 to 15. A computer program product, wherein the computer program product is stored in a non-volatile, readable storage medium, and the computer program product is executed by at least one processor to perform the point cloud attribute coding of any one of claims 1 to 5. A computer program product implementing the steps of the method, or implementing the steps of the point cloud attribute decoding method of any one of claims 6 to 15. 16. A communication device, configured to perform the steps of the point cloud attribute coding method of any one of claims 1 to 5, or to perform the steps of the point cloud attribute decoding method of any of claims 6 to 15. A communication device configured to